Joint Communication and Processor Frequency Selection for Low-Energy Systems under Timing Constraints

This paper presents a method to reduce the energy consumption of multi-core systems characterized by processor cores and buses with discrete frequency levels under timing constraints.The proposed method takes the transformations of the original task graphs,which include dependent tasks located in different iterations,as inputs.The proposed method utilizes mapping selection as well as joint processor and communication frequency scaling to implement energy reduction.We conduct experiments on several random task graphs.Experimental results show that the proposed method can achieve substantial energy reduction compared with previous work under the same hard timing constraints.

Adaptive Unified Biased Estimators of Parameters in Linear Model

To tackle multi collinearity or ill-conditioned design matrices in linear models, adaptive biased estimators such as the time-honored Stein estimator, the ridge and the principal component estimators have been studied intensively. To study when a biased estimator uniformly outperforms the least squares estimator, some sufficient conditions are proposed in the literature. In this paper, we propose a unified framework to formulate a class of adaptive biased estimators. This class includes all existing biased estimators and some new ones. A sufficient condition for outperforming the least squares estimator is proposed. In terms of selecting parameters in the condition, we can obtain all double-type conditions in the literature.

Bayesian doubly adaptive randomization in clinical trials

Bayesian adaptive randomization has attracted increasingly attention in the literature and has been implemented in many phase II clinical trials. Doubly adaptive biased coin design(DBCD) is a superior choice in response-adaptive designs owing to its promising properties. In this paper, we propose a randomized design by combining Bayesian adaptive randomization with doubly adaptive biased coin design. By selecting a fixed tuning parameter, the proposed randomization procedure can target an explicit allocation proportion, and assign more patients to the better treatment simultaneously. Moreover, the proposed randomization is efficient to detect treatment differences. We illustrate the proposed design by its applications to both discrete and continuous responses, and evaluate its operating features through simulation studies.

A novel broadband power amplifier in SiGe HBT technology

A novel broadband power amplifier fabricated in 0.13 m SiGe HBT technology is realized.The pseudo-differential structure is proposed to avoid the influence of the bonding wire due to the AC virtual ground created at the common emitter node.A compensated matching technique is adopted in interstage matching to expand bandwidth.A multi-stage broadband matching technique is used in an input/output matching network to offer broadband impedance matching,which ensures maximum power transfer.An adaptive bias circuit could improve linearity and efficiency in wide output power level.With 2.5 V power supply,the measured results achieve 96% 3-dB bandwidth（517-1470 MHz）,27.2 dB power gain,26.9 dBm maximum output power,19.7 dBm output 1 dB compression point,and 26.7% power added efficiency.

A highly linear power amplifier for WLAN

A three-stage power amplifier （PA） for WLAN application in 2.4-2.5 GHz is presented. The proposed PA employs an adaptive bias circuit to adjust the operating point of the PA to improve the linearity of the PA. Two methods to short the 2nd harmonic circuit are compared in the area of efficiency and gain of the PA. The PA is taped out in the process of 2/zm InGaP/GaAs HBT and is tested by the evaluation board. The measured results show that 31.5 dB power gain and 29.3 dBm PldB with an associated 40.4% power added efficiency （PAE） under the single tone stimulus. Up to 26.5 dBm output power can be achieved with an error vector magnitude （EVM） of lower than 3% under the 64QAM/OFDM WLAN stimulus.

Theoretical analysis and an improvement method of the bias effect on the linearity of RF linear power amplifiers

Based on the Gummel-Poon model of BJT, the change of the DC bias as a function of the AC input signal in RF linear power amplifiers is theoretically derived, so that the linearity of different DC bias circuits can be interpreted and compared. According to the analysis results, a quantitative adaptive DC bias circuit is proposed, which can improve the linearity and efficiency. From the simulation and test results, we draw conclusions on how to improve the design of linear power amplifier.